Multiple-choice questions on PathMMU PathCLS n = 177 (test-tiny)
78.9AccuracyExpert performance
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Expert performance2025.06 | 78.9 | — | |
| PathChat+2025.06 | 74 | 67.8 | |
| GPT-52025.06 | 53.7 | 46.3 | |
| Gemini 2.5 pro2025.06 | 52.5 | 45.8 | |
| GPT-5-mini2025.06 | 50.3 | 42.9 | |
| PathChat 12025.06 | 42.4 | 35.6 | |
| Qwen3-VL2025.06 | 38.4 | 31.6 | |
| HuatuoGPT-Vision2025.06 | 36.7 | 29.9 | |
| Claude Sonnet 42025.06 | 35 | 28.8 | |
| Llama-3.22025.06 | 31.1 | 24.9 | |
| PA-LLaVA2025.06 | 30.5 | 23.7 | |
| Quilt-LLaVA2025.06 | 29.9 | 23.2 | |
| LLaVA-OneVision2025.06 | 22.6 | 16.4 |